Links For May


[Remember, I haven’t independently verified each link. On average, commenters will end up spotting evidence that around two or three of the links in each links post are wrong or misleading. I correct these as I see them, and will highlight important corrections later, but I can’t guarantee I will have caught them all by the time you read this.]

1: Apparently one important step on the way to healing partisan divides in America is implementing prophecy reform. “Yes, prophecy reform.”

2: For the first time since 1797, someone has used the infamous Venetian doge selection process to select an officeholder - specifically, the new moderators of not-quite-officially-affiliated-with-ACX politics discussion subreddit r/TheMotte. I assume this is why dogecoin is up this month.

3: Best of Less Wrong: Seven Years Of Spaced Repetition Software In The Classroom. Describes a teacher’s experiments with Anki / Supermemo style SRS flashcards; the conclusion is that using them is complicated, they sort of work, but they helped him realize how much of learning isn’t about memorizing things. I appreciated this most for its theory that it’s important to make kids learn specific facts, but not so important that they remember them; teaching someone (eg) Civil War history is “training” a “predictive model” of the Civil War, war in general, and history in general which will survive and remain useful even after the specific facts and battles are long forgotten. I think this is the strongest defense of modern education, given that we do spend lots of time teaching kids things they will definitely forget. But how would you test it?

4: I’m very late here, but you might still enjoy OpenAI’s Jukebox, which is basically GPT-3 for music. Train it on Elvis, then make it write new songs on his style. Or feed it the first few verses of Never Gonna Give You Up and make it guess what the rest of the song sounds like. Or just have Celine Dion sing a song about being a music generation algorithm produced by OpenAI.

5: Did you know: the first President of Zimbabwe was named Canaan Banana.


See here for more discussion by the author.

7: Best of Less Wrong: DARPA Digital Tutor: Four Months To Total Technical Expertise? In 2009, DARPA created a digital tutoring system that could adjust lessons based on students’ strong and weak points. After four months, digitally-tutored IT technicians outperformed experienced professionals in DARPA’s tests. How is this different from existing digital learning software, and could we make equally successful programs for other subjects?

8: Related? Given tablets but no teachers, Ethiopian children teach themselves. But see this comment for reasons to be skeptical.

9: A journalist asked New York mayor candidates to estimate the median price of a house in Brooklyn. Guesses were 90K, 100K, 500K, 550K, 800K, 900K, 1M, and 1.8M; the correct answer was 900K. The two candidates who guessed 90K and 100K are under fire for being spectacularly out of touch (one of them was a former US Cabinet Secretary of Housing and Urban Development). Usually I am at least sort of sympathetic to out-of-touch politicians, because I myself am out-of-touch, but I can’t imagine making this particular error. Also, Andrew Yang was the guy who got it exactly right; he claims the number just “popped into my mind”.

10: A fun perspective on the typical image used to prove something about American Car Culture or Capitalism or Whatever:

Apparently it took the photographer three days of trying before getting a shot that looked that bad. And also, the “town” was built at a unique area where a quirk of federal highway funding banned two highways from connecting directly without a stretch of normal small-town road in-between; since the small-town road was just a cover for the major highway connection, of course it ended up as very car-centric. Probably you should not use this as evidence of anything.

11: Did you know: Leon Trotsky’s great-granddaughter, who grew up in the Mexico City house where he was assassinated, is now one of America’s top psychiatrists and director of the National Institute on Drug Abuse. I feel like there’s a “drug czar” joke to be made somewhere here.

12: LanguageLog: How does naming new chemical elements work in Chinese? Answer: you have very short words for each element, vaguely based on the Western name - for example, aluminum is “lǚ” and rutherfordium is “lú” - the characters are all the character for “metal” or “gas” or something plus something else - and it becomes so confusing that a commenter speculates it might be significantly holding back China’s technological progress.

13: Police in Poland have detained seven people on charges relating to the building an enormous medieval-style castle on a lake in Notecka Forest, an area protected as part of the EU’s Natura 2000 network.”

Here in Oakland we also sometimes detain people for unapproved building projects, but it’s usually more like “tried to repair their doorknob without getting enough permits”, so it’s hard for me to feel anything other than impressed by the Poles for making it this far.

14: Related: did you know: far-right anti-immigrant group VDare owns an honest-to-goodness castle. On reflection I am okay with this; castles are a traditional form of villain lair, and thematically appropriate for white nationalism.

15: Tired of being outdone by all those politicians playing 4D chess? Now you can play 5D Chess With Multiverse Time Travel.

16: Okay, sure, I’m just including this whole tweet:

17: Best of Less Wrong: Are We In An AI Overhang? IE a situation where we have almost all the pieces we need to make much smarter AIs than we’re currently making, and once we snap the last piece into place everything will start moving really fast.

18: Bret Deveraux talks about the sources of cost disease in universities; he suggests “the bloat” comes from a new layer of “vice-deans” and “vice-provosts” and various attempts to centralize administration in a way that just creates a duplicate and worse administration beside the old decentralized one.

19: In response to a question about hype in media, AOC gives a pretty scary story about how NBC deliberately misinterpreted her actions at the Democratic convention.

20: What jobs are most disproportionately transmitted from parents to kids? (here’s the source, with more information, but it’s paywalled) In general, people who have a parent with job X are about twice as likely to have job X as a random member of the population. But programmers are 6x more likely, lawyers are 18x more likely, doctors 25x more likely, dentists 80x more likely, and journalists 94x more likely. But the most transmissible jobs all seem blue-collar, with the overall winner being textile machine operators (415x more likely, if both parent and child are male).

21: Quanta: What Sonic Black Holes Say About Real Ones

22: I learned about life history strategy theory from Evolutionary Psychopathology, which I reviewed here. But a recent study concludes there is “little evidence that humans follow coherent life-history strategies”.

23: Another tweet-story I’m just going to quote in full, this one courtesy of Trump staffer Cliff Sims:

24: Fact-checkers take on the Babylon Bee’s claim that the Ninth Circuit Court of Appeals overturned the death of Ruth Bader-Ginsberg, for some reason.


26: Can you tell GPT-2-generated scientific paper abstracts from the real thing? I’m pretty proud of my 90% success rate on hard mode.

27: Harvard Business Review on when corporate philanthropy is good vs. cringe. My favorite sentence: "Tobacco giant Philip Morris…spent $ 75 million on its charitable contributions in 1999 and then launched a $ 100 million advertising campaign to publicize them."

28: Very thorough explanation of why the Moscow Metro is so beautiful, and how it relates to the usually-more-utilitarian Soviet aesthetic.

29: In 82 BC, the dictator Sulla executed scholar-politician Quintus Valerius Soranus for publicly revealing the secret true name of Rome. Wait, there was a secret true name of Rome? Apparently yes, and it had to never be revealed, lest Rome’s enemies learn it and gain spiritual power over the city. The execution must have worked, because scholars today still don’t know what Rome’s secret true name was. Irresponsible speculation includes Maia (via astronomical correspondence), Amor (via wordplay), Valentia (via multilingual pun), and Hirpa (via a long chain of scholarly/historical speculation).

30: Related:

31: Via Siberian Fox, one reason evolution might not have eliminated the appendix even if it’s useless:

32: Best of Less Wrong: Optimized Propaganda With Bayesian Networks. If you understand exactly why people believe something, and how much each prerequisite belief contributes to the final product, you can determine where to “push” to change minds fastest. Not just in a theoretical way, in a “somebody actually mapped this out for anti-vaccination beliefs, numbers and all” way. This is potentially good if you want to change people’s minds, potentially bad if you are a person with a belief who doesn’t want propagandists and marketers getting cool algorithms about how to change your mind about it most effectively.

33: I’m enjoying the mathematical art of Rafael Araujo, eg:

34: Also David Li, who does interactive mathematical program/art which I can’t embed here, like this sphere vortex (try moving your mouse around the page)

35: Recent news in local AI alignment research space: most of OpenAI’s top alignment researchers, including Dario Amodei, Chris Olah, Jack Clark, and Paul Christano, left en masse for poorly-understood reasons (see speculation here). Dario Amodei is now working with a new nonprofit called Cooperative AI Foundation. Paul Christiano will be founding his own nonprofit, the Alignment Research Center (conflict of interest notice: I know Paul and think he is generally great); see also his ask-me-anything thread on Less Wrong here. Unrelatedly, local secretive AI alignment research group MIRI (Machine Intelligence Research Institute) is leaving the Bay Area for some small town with affordable land prices where they can maybe build a campus (they’re still trying to decide exactly where).

36: Related: Daniel Filan now has an AI X-Risk Research Podcast (conflict of interest notice: I know Daniel and think he is generally great).

37: CelestialSEM is a Twitter account with only one gag, but it’s a good gag: structural equation models look funny if you give them a starry background and an outer glow: